Determining Overlap Percentage in Drone Collected Imagery
New in Global Mapper v24.0 is the ability to measure image overlap with the Image Overlap Report in our drone imagery processing tool, Pixels to Points. Overlap between images is required in photogrammetry in order to provide enough data for the software to extract positioning information. An object or feature needs to be captured in at least three images for it to have enough information/data to be generated. For more information on how photogrammetry and Pixels to Points work, check out the second section of this YouTube video on Structure from Motion.
Image overlap is something that is taken into account during flight planning, but assessing this prior to processing can help identify areas in the data that are weak and can be omitted. The recommended overlap percentage is 60% front and 40% side. More is fine, but less can lead to holes in your final generated outputs. You can see the effects of low overlap at the periphery of most photogrammetrically generated data. By nature, the edges have less overlap, therefore, less data, so the outputs are more likely to be sparse and less accurate in those areas. Measuring overlap can highlight weaknesses in the data set, and help create a higher quality output.
The Image Overlap Report automatically generates when you create Ground Coverage Polygons (also referred to in the industry as “footprints”). The Overlap Report can be generated for individual images, or for all images.
First, load the images into the Pixels to Points (P2P) tool, select the image(s) you would like to measure the overlap of, then right-click and choose “Load Ground Coverage Polygons for Selected images.” The coverage polygons will be generated in the workspace, and the Overlap Report will open in your default browser.
Note that before the images have been processed through Pixels to Points, the overlap percentages are based on an estimation* of the footprints. This means that the footprints generated before being processed by Pixels to Points will be different then those generated after the reconstruction has finished. The difference is because the Pixels to Points reconstruction process will refine the camera parameters and position.
*Why are the footprints estimated prior to running Pixels to Points? When a UAV takes a picture, it typically only records the XYZ position of the center of the image. From this, Global Mapper knows where the image is located, but it doesn’t know the exact orientation or size of the image. Image footprints are officially calculated in Pixels to Points when you click Run. They can also be seen without Pixels to Points, as the footprints are quickly estimated based on 3D location, recognized camera make/model, and camera/gimbal orientation.
Similar to the Processing Report, the Overlap Report will open in your default browser. It is saved in a temporary folder on your machine. To save it for future use, or to send it to another user, you can save it through your browser (ex. Print to PDF).
The Overlap Report has four sections: Overlap Statistics Misc, Image Relative Areas, Hit Overlap Histogram, and Side Neighbor Images. The final three sections describe the error of the statistics used to estimate the footprints, how much area is covered by how many images, and a list of overlap. For more information on these three, see the Knowledge Base.
The first section, Overlap Statistics Misc, is likely the most important and the only area most users will need to review. This table shows you the overlap statistics for the processed image(s):
The Overlap Statistics table displays the overlap percentages and a few estimates of accuracy for the footprint measurements.
“Overlap Ratio” – is the amount of area that is covered by at least 2 images. The area is estimated by a bounding box. This value may be low if your images aren’t in a rectangular pattern.
“Average Area Calculation Error” – It is easy to calculate an area value for a rectangular area. However, calculating the area of intersection between several rectangles can be very time-consuming. As a shortcut, Global Mapper uses approximation approach, and its error is shown here. A 1% error means we calculated 101 instead of 100. If the error is high, try generating the footprints again after the Pixels to Points reconstruction process.
“Forward Disconnect” – is a counter of non-overlapping images in the forward flight direction. The forward direction is defined by the timestamps of the images.
For more information on the other sections of the Overlap report, please consult the Knowledge Base.
For blog topic requests, please email email@example.com